1. Field of the Invention
The present invention relates to data communications, and, in particular, to equalization of a signal through a communications channel.
2. Background of the Invention
In many data communication applications, serializer and de-serializer (SerDes) devices facilitate the transmission between two points of parallel data across a serial link. Data at one point is converted from parallel data to serial data and transmitted through a communications channel to the second point where it received and converted from serial data to parallel data.
At high data rates frequency-dependent signal loss from the communications channel (the signal path between the two end points of a serial link), as well as signal dispersion and distortion, can occur. As such, the communications channel, whether wired, optical, or wireless, acts as a filter and might be modeled in the frequency domain with a transfer function. Correction for frequency dependent losses of the communications channel, and other forms of signal degradation, often require signal equalization at a receiver of the signal. Equalization through use of one or more equalizers compensates for the signal degradation to improve communication quality. Equalization may also be employed at the transmit side to pre-condition the signal. Equalization, a form of filtering, generally requires some estimate of the transfer function of the channel to set its filter parameters.
In many cases, the specific frequency-dependent signal degradation characteristics of a communications channel are unknown, and often vary with time. In such cases, an equalizer with adaptive setting of parameters providing sufficient adjustable range might be employed to mitigate the signal degradation of the signal transmitted through the communications channel. An automatic adaptation process is often employed to adjust the equalizer's response.
In practical implementations of the adaptation processes, variants of least mean square (LMS) adaptation might be used for setting values of equalizer parameter. Values such as, for example, feedback post cursor and feed-forward finite impulse response (FIR) taps in a digital filter, or pole and zero values for an analog filter, equalizer implementation are calculated by optimizing a LMS cost-function based on observation of the received signal over time. Some classical adaptation schemes estimate the channel impulse response by optimizing the minimum mean-squared error cost function between the desired signal, d(n), and the equalized signal, q(n).
After channel estimation, the contributions of inter symbol interference (ISI) due to past detected symbols might be removed from the receiver input signal using Decision Feedback Equalization (DFE). In doing so, the ISI signal spreading is reduced towards an optimal point. In this case, a decision, y(n), is generated and equalized used to provide the desired signal, d(n). Using the LMS adaptation algorithm and sampler array, a receiver calculates a decision error, ε(n), as in equation (1):ε(n)=d(n)−q(n),  (1)and then sets the equalizer parameter values so as to minimize this decision error, ε(n), by optimizing the parameter values under some criterion. Classical adaptive filters minimize the mean square error of ε(n) to achieve the adapted filter parameter values such as, for example, filter tap coefficients of a DFE or pole/zero locations of an analog filter. When optimal filter parameter values are approximately achieved, the derivative of the mean square error with respect to the filter coefficients is zero.
FIG. 1 shows a block diagram of a prior art adaptive equalizer 100. Equalizer 100 comprises DFE 101, sampler array 102, and DFE tap generator 103. Input samples, x(n), are equalized by combination of i) x(n) and ii) decision feedback equalizer error correction signal, e(n), in combiner 110 of DFE 101 to generate an equalized signal, q(n). Sampler array 102 includes top error sampler 111, data sampler 112, and bottom error sampler 113. Top error sampler 111, data sampler 112, and bottom error sampler 113 might be implemented as simple slicers, or as a threshold comparators and latches. The equalized signal, q(n), is sampled by data sampler 112 to generate a decision, y(n), which is also the data output signal.
FIG. 2 shows a data eye diagram 200 overlaid with exemplary data sampler 112 and error samplers 111 and 113. In many practical implementations, the error signal, ε(n), is obtained by placing top error sampler 111 and bottom error sampler 113 at the top eye edge 203 and the bottom eye edge 204 of the data eye. Ideally, the sampler reference voltages of the top error sampler 111 and the bottom error sampler 113 are set at the threshold voltage level that would be achieved after a perfect equalization. Since information of the threshold voltage level that would be achieved for perfect equalization is not known until equalization is applied, an RMS value of the signal is used to estimate the sampler threshold voltage level.
Classical adaptive filters minimize the mean square error of ε(n) to achieve the adapted filter parameter values such as, for example, filter tap coefficients of a DFE or pole/zero locations of an analog filter. When optimal filter parameter values are achieved, the derivative of the mean square error with respect to the filter coefficients will tend to be zero.